A Research Framework for the Archaeology of Wales

East and Northeast Wales – Roman




Our understanding of this, the largest of the four regions into which Wales has been divided for the purpose of conducting the Regional Research Assessment Seminars, negates a current perception of the Romano-British period as over-researched, the exercise being essentially repetitive, even sterile.  In the Romano-British period it is probably the Cinderella of the Welsh regions, with yawning gaps in our very basic knowledge.  Indeed, as I hope to demonstrate, without a substantial increase in the volume and quality of our data-base we cannot even begin to address the more challenging questions pertaining to the effects of incorporation into the Empire   -as outlined in James & Millett’s recently published edited volume (essentially pertaining to England) Britons and Romans: advancing an archaeological agenda (CBA: York: 2001).   A comparable exercise for Wales as a whole (though such geographical divisions in the Roman age is entirely artificial and thus undesirable) can only be pursued if the above criteria are satisfied.  If not, then such an exercise becomes largely imitative and highly theoretical in character.


Our region is characterised by a lack of obvious signs of Romanity in its settlement pattern; urban centres, of civitas type or of the ‘small-town’ variety, are absent.  Only one site can securely claim villa status.  Military remains are all too obvious, testifying to a seemingly protracted presence.  In the pre-Flavian period, with Gwent and the Marches, it was the scene of intense military activity.  It is thus one of the most interesting areas of study for a number of reasons.  Lying on the interface between the West Midlands and the mountainous heart of Wales it spanned a strategic interface for the military, whilst at the same it has a propensity to produce highly distinctive settlement forms exhibiting the characteristics of both the English lowlands and Welsh uplands.


Interfaces have always proved attractive from a research standpoint, and our region is no exception.  There have been numerous publications on the Welsh Borderland, though the Welsh side of the political divide has proved the poor relation. At the same time we need to bear in mind that we cannot understand developments within our region without reference to other parts of Wales, the neighbouring English Counties and, of course, Britannia as a whole.  In this respect it is necessary to appreciate the fact that the Roman age was far from static, least of all at a socio-political level.  For example, by the third century there were far-reaching changes in the system of taxation; new social hierarchies were emerging out of earlier systems; military frameworks too were being altered, and both military and political structures had been fundamentally changed by the time that Wales formed part of Britannia Prima in the fourth century.


My brief is to firstly, outline what the Resource Audit for our region informs us about the region within the context of a SWOT analysis based upon some traditional categories of site and artefact, falling within the compass of the broad themes of Invasion and Military Occupation, Romanization/Acculturation and Exploitation.  Secondly, to consider research strategies for the furtherance of our knowledge concerning the region in the Roman period, including the addressing of new questions arising from research issues in an over-arching Roman dimension.





Many military sites, belonging to both the campaigning and the subsequent garrisoning phases are known, and may be contextualised with reference to like sites in Wales and the Marches as well as Roman Britain in general.


The region has the largest number of marching-camps in Wales, some of which can be tentatively linked with historically attested operations initiated beyond the present day political boundary.


With Gwent the region also has the greatest potential for understanding the course of military operations in the pre-Flavian era (i.e. pre-AD 73/4).  Sites such as Clyro, and Rhyn (both hard on the border) are invaluable sources for the understanding the course of early campaigning. In this context there are also a number of enigmatic, possibly unique military sites in the Llanymynech area.


The conquest and pacification of the region in the early Flavian period is well established. It is probable that the majority of the auxiliary forts founded in the Flavian period are known to us, though to judge by an analysis of the road network, and the dearth of forts in NE Wales, some either remain undiscovered (in the valley of the Banwy or the Vyrnwy) or a fresh appraisal of the distribution of military posts is required.  The situation in eastern Powys in particular needs to be clarified.


All known auxiliary forts have been subject to some excavation; some such as Brecon Gaer, Caersws or Castell Collen on a large scale.  Others, such as Caerau, Pen y Gaer, Cae Gaer, Hindwell Farm and Colwyn Castle have received only the most cursory treatment.  The broad chronology of this class of monument is nevertheless well established, but even in the context of those well-explored forts changes in their internal layout and the date of the refurbishment of defences and replacement of buildings in stone are by no means firmly dated, whilst the reasons for such in a site-specific, as well as a wider military context are either unclear or unknown.


Associated military vici are known at a number of fort sites, though it is only at Caersws that we can claim sound knowledge of their plan and overall history, due to recurrent, recent and indeed continuing excavations.  Caersws is one of the most intensively excavated military vici in Britain, and we know more about its chronology. Plan and economic activities focused upon it, than any comparable site in Wales.


On Llandrindod Common we have the largest cluster of practice-camps in Wales (and in Britain as a whole).


At Holt we have an exceptionally important complex serving the needs of the legionary fortress at Chester insofar as brick/tile, and to a lesser extent pottery, is concerned.  Its remains are hitherto unique. Though excavations were undertaken long ago, re-assessment of its chronology and projected geophysical survey will enhance our knowledge of this site.



The chronology of the great majority of those sites connected with the pre-Flavian and Flavian campaigning phases is imperfect.  This is certainly the case with the marching-camps (not an unusual state of affairs), but of more concern in the case of campaigning bases such as Clyro or forts such as Hindwell Farm.


The number of known pre-Flavian forts is very small.  Does this reflect the true state of affairs or not?  Those that are known cannot be presently fitted into a chronologically sound pattern of military activity. Are there pre- Flavian and Flavian forts to be discovered in the Clwyd valley and in NE Wales in general?  There are tentative hints of such at Ruthin and Rhuddlan, but the matter remains to be resolved.


Some so-called military sites  - especially some of those so categorized by the late G.D.B. Jones-   may ultimately prove to be illusory.  For example, sites in the Abertanat/Clawdd Coch complex near Llanymynech, the so-called stores-base at Llansantffraid ym Mechain and the ‘forts’ identified by him at Prestatyn and Ruthin.


Our knowledge of some of the basic features of the Flavian and later military posts  -forts and fortlets- is extremely poor, and largely based upon excavations undertaken up to half a century ago; in some cases, as at Forden, even longer.  Small-scale excavations have also, for the most part, only produced the most generalized information as to their overall chronology, whilst their internal layout is sketchy in the extreme.  Virtually nothing is known of fort plans of the Flavian age. There is also circumstantial evidence that some of the earliest Flavian forts may have been large, and were subsequently abandoned in favour of standard types of garrison bases (for example; Llwyn y Brain and Caersws). Such a tentative sequence needs to be investigated. 


For some sites  - Cae Gaer and the fortlets in the region- there are scarcely any chronological indicators at all.


Whilst small-scale excavations have produced snapshot chronologies  -shaky in the extreme, and conceivably faulty- they have proved incapable of addressing the more challenging issues surrounding the history of the sites in question.  Were they ever fully occupied?  Were forts held by units themselves divided between several forts?


Did those forts which produce evidence of occupation in the mid-third to later fourth century accommodate full-sized units in this period?   Could there have been alternative uses for such sites as police posts or repositories for local tax-dues?  Why, for example, was the long-abandoned fort at Forden re-occupied in the Valentinianic era?


Is the site classified as a signal-station to the west of Hindwell Farm fort a Roman military installation?


Caersws and Castell Collen excepted, the extent, morphology and chronology of associated military vici is far from clear.  There are chronological hints as to the longevity of the vici at Forden and Brecon Gaer, but no more.  Is there even a third-early fourth century vicus at Caersws?  Such a dearth of basic information pertaining to the growth, contraction and abandonment of vici severely hinders any assessment of the relationship between military-related settlement and civilian communities at large.


Is there any evidence of continuity of settlement on a fort/vicus site into the late-Roman/Early Christian period?



To further examine the vicus at Caersws as housing infill continues.


To elucidate the sequence at Forden in response to the long-term threat of plough damage to this remarkable complex.  The existence of presumed pre Roman Iron Age and ?Early Medieval settlement in the immediate vicinity makes this an ideal site for a long-term research project.





The strategic road system, intimately connected with the needs of military supply has been extensively researched thanks to the fieldwork of individuals such as W. Putnam, Prof G.D.B Jones and H. Toller and the course of many of the roads is now securely established.



In certain areas there is little firm evidence as to the routes taken by the roads.  For example, Chester – Caerhun; Chester – Caer Gai; Caer Gai – Vyrnwy/Banwy area; the Wye valley/Usk valley interface and the presumed Vale of Clwyd road.


There are a number of alternative routes, some probably fanciful, in some of the above areas but the ‘Roman Roads’ project currently under way by CPAT will lead to a much better understanding of the system and an infilling of those blank or speculative areas where roads are deemed to have existed but the routes not yet determined.


Virtually nothing known of the bridges spanning the major rivers of the region.


Nothing known of other communication systems, such as sea-borne traffic and the use made of navigable rivers.



Considerable interest in the investigation of Roman roads, with both amateurs and professionals engaged in research, utilising a variety of source material, including antiquarian tradition, place-name evidence and data from aerial photographs.



Lack of collaborative effort and exchange of information.




Since our region is bereft of towns all non-military settlements fall within the compass of ‘Rural Settlement’, though there is a manifest distinction between those which appear to reflect continuity  -at least in terms of site location- from the pre-Roman Iron Age and those which are manifestly products of Roman acculturation and sometimes frequently associated with economic exploitation.



The recognition of settlements occupied in the RB period through the existence of a commonly found range of reasonably well-dated artefacts.


Those sites that have been explored of late are all published.  For example: Dinorben (H N Savory’s work), The Breiddin, Collfryn, Rhuddlan, Pentre (Flint), Prestatyn, Maesderwen and Broad Heath.  There is a good publication record in the region.


Those that have been excavated have produced a valuable range of environmental evidence, though the non-survival of animal bones is a problem in some parts of the region.


There is a recognisable diversity of site-type, ranging from a probable villa (Maesderwen) to hillforts, to a variety of small, ditched and embanked enclosures to caves.  Such suggest a settlement hierarchy and differences in economy as well as status.


Work at Plas Coch (Wrexham) illustrates something if the nature and settlement characteristics of the lowland RB landscape in our area, without having to focus on a site-specific context.


The tendency for some sites to have a pre-Roman phase allows the recovery of data pertaining to such issues as settlement continuity and the social and economic impact of conquest etc.  However, some sites, such as the enclosures investigated in the Walton Basin, seem to have been abandoned before, or upon the conquest.  Or did they continue to be occupied without taking up clear indices of Roman acculturation?



It is considerably easier to discuss the weaknesses of the resource for this region than its strengths.


The total number of sites known to have been occupied in the RB period is tiny. Indeed the proportion is the lowest in Wales.  Hillforts excepted, the evidence is also heavily biased towards the Borderland and to enclosures in lowland settings.  Large areas  -Powys in particular- are virtually devoid of evidence for settlement, even in those areas where the density of pre-Roman Iron Age sites is such as to suggest that occupation in the Roman period is inherently likely.


It has been virtually impossible to discuss issues such as how settlement history and economic practices were effected in the Roman period because of the exceedingly small database.  Similarly issues such as continuity/discontinuity, the role of elites and the growth of a market economy are difficult to debate.  Even data pertaining to the retention of customary architectural forms, is small.


The Resource Audit lists 7 villa-type settlements in the region, though this appellation has been extended to administrative-type buildings such as that at Pentre (Flint).  Even the exact status of the only building complex that has all the hallmarks of a late Roman villa  -Maesderwen- needs to be proved beyond doubt.


A number of hill-forts, particularly in the northern portion of our region, show evidence of RB activity though its character is frequently unclear.

What is their relationship to non hill-fort settlement in this period?


Cave sites are again little understood.  Is the occupation domestic?  The material assemblages often suggest that ritual usage is more likely.  Clarification is necessary here.


Scarcely anything is known about land-use in general, though aerial photography and a little surface evidence shows traces of fields/enclosures in the vicinity of settlements such as Collfryn.  The chronology of such remains, however, is frequently uncertain.


There is a general dearth of information concerning environmental issues in general.  Animal bone and crop assemblages are scarce; hence a lack of some very basic comparative information on subsistence strategies.  How may we begin to address issues such as to whether garrison posts in the region drew their supplies from local resources, or changes in the economic basis of contemporary settlements if we lack the basic environmental data?


There is a variety of scattered artifactual evidence such as coins (including hoards), pottery and sundry items of metalwork that testify to the introduction of Romanised culture into a rural environment.  But how are these to be contextualized?  The full impact of changes in material culture brought about by a market economy is presently beyond our grasp.  How may elite/peasant relationships be expressed in such terms?  How do we go about challenging traditional perceptions/narratives, such as Romanization, without good quality evidence?  In this respect it is possible that the Portable Antiquities Scheme may prove valuable in providing better information about the distribution of finds in general within the region.



Opportunities for future research are most likely to be limited and confined to those areas where development-led threats are most likely, such as the vicinity of towns and villages; and in the NE part of the region in particular.  However, the recent example of rescue excavation at Arddleen illustrates the haphazard nature of such threats’ Here it will facilitate the complete excavation of a lowland enclosure known to have produced RB material.


Practically any site in the region defended or otherwise, categorised as being of ‘Iron Age’ date, for want of a better term, has the potential to demonstrate a RB phase in its history.  Thus, any threat, involving excavation on such a site presents an opportunity to peer through a window into the Romano-British period.


Whenever possible the opportunity be taken to seek out traces of RB settlement in those areas where such is a virtual blank; utilising aerial photography and field survey, especially in those lowland areas such the middle Usk and upper Wye valleys where such may be expected.  Aerial photography may demonstrate the existence of other villa-type settlements in such areas, and as such a programme of this kind should be accorded a high priority.  Funding should be adequate to cover the needs of more aerial reconnaissance over these zones in bumper years.  Similarly, geophysics needs to be employed to sample the interior of some enclosures in expectation of the recovery of Romanised types of building plans together with anomalies such as may be caused by the presence of corn-driers, a frequently occurring structural component of lowland farmsteads in the RB period.


In the case of settlements of overtly Romanised character, such as Pentre (Flint), Prestatyn, Ffrith and Plas Coch the full extent of these complexes needs to be determined through geophysical survey


Environmental sampling and analysis of promising peat/mire deposits, together with any opportunities offered for the sampling of field/enclosure systems should be exploited. The investigation of coastal peat deposits has much to contribute to our understanding of the settlement pattern and landscape of the coastal strip between the Dee and the Conwy.


Opportunities for the investigation of RB activity within hillforts is likely to recur from time to time.  Large-scale work within sites being destroyed through quarrying, such as The Breiddin, could produce vital information.



The possibility that fresh information is only likely to come from those areas under pressure from urban expansion.  Elsewhere the small-scale development arising from economic development is only likely to lead to fragmentary advances in knowledge, unless the nettle is grasped and the excavation of plough-damaged sites (sites categorized as under normal agricultural regimes) is undertaken on a regular, but selective basis.  Otherwise there is a clear threat of stagnation in our knowledge.


The possibility that there may be no large-scale excavations in the near future.


Inadequate funding, particularly of aerial reconnaissance is likely to remain a problem


The continued ploughing of low profile embanked and ditched enclosures and sites reduced to cropmark status, is likely to lead to the continued degradation of this resource in archaeological terms.





The RB period saw the introduction of new technologies and crafts, and most significantly a fundamental change in the scale of production and exchange.


Some of the sites in our region have a high potential for the elucidation of industrial processes, particularly those associated with the exploitation of metalliferous deposits.  Indeed, we have more known, or suspected sites concerned with industrial processes in our region than any other part of Wales.  For example, those connected with lead-silver extraction along the NE seaboard.  Hard on the border at Llanymynech and in western Montgomeryshire we have other locations where lead-silver and copper may have been extracted.  Llanymynech in particular has high potential, since its ores were certainly being exploited in the pre-Roman period, and the complex may be particularly instructive in respect of continuity of exploitation.


We know something of the administrative arrangements and the chronology of lead-silver exploitation on the basis of Imperial stamped pigs, and from the exploration of buildings such as the ‘villa’ at Pentre in the associated mining settlements, together with arrangements for the shipment of the metals.


We have the rare evidence of a workshop producing bronze brooches and other personalia at Prestatyn.  Whether this is native opportunism is unclear.


Native (?) iron smelting is known at Maesderwen and Gwernyfed Park in Powys.


Brick/tile production, and pottery manufacture, is known at Holt and Caersws, with kilns for the production of the former at both sites.


Water technology is demonstrated by the ubiquity of wells in fort/vicus contexts, whilst the region has the only excavated elements of a timber aqueduct in Wales at Prestatyn.



Insofar as metal extraction is concerned the extraction sites are unexplored, and much interfered with by later workings.  Research has been largely focussed upon earlier prehistoric exploitation


The excavation of processing sites and the associated infrastructure has been largely small-scale.


The context of some metalworking is unclear.  Are the furnaces at Gwernyfed Park civilian or military in origin?  Was iron in billet form largely imported into the region or did native communities play a key role in the exploitation of local resources?


The examination of pottery assemblages from military contexts (for example, Caersws) makes it clear that local kilns were operating within the region in the Flavian-Trajanic period. None of the production sites have been identified to date, though they are likely to lie in close proximity to garrison bases.  Did the products of these relatively short-lived enterprises get into native hands?


Our knowledge of the sourcing and the usage of ceramics on rural settlements is extremely poor.


Whilst small-scale industrial/craft activity is known in vicus-type contexts, virtually nothing is known about those in purely native hands.  Was craft-activity centralized or not?  What was its scale?


We can demonstrate the exploitation of the lithic resources of the region, and the ‘importation’ of stone from outside, as in the case of the slate tombstones from a cemetery near Brecon Gaer; and even hazard the origin of the stone utilised in building the fort walls at Caersws. But very little work has been done on the trade in manufactured goods, such as quernstones, or on the identification of quarries.


What of tanning/curing of hides or the dying of cloth?  Nothing is known of this activity, in either military or civil contexts.  Could the examination of palaeoenvironmental deposits in excavated contexts throw light on this activity?


We would like to know whether the development of a transport infrastructure and the need for military supply affected the distribution of traded commodities?  It is clear that military and military-related sites produce different artefact assemblages in some areas.  Is this also true of our region?  We need to know what access native communities had to the market; whether this pattern differed within our region; and, how did it compare with other parts of Wales.



Continued academic interest in artefact assemblages and in the distribution of finished goods and consumables will lead to the comparative analysis of the incidence and distribution of material within the region.


Continued interest in early mineral extraction, and in mining sites in general, continues at both amateur and professional level.  The enthusiasm of the former can produce break-throughs in our knowledge, as illustrated in Early Bronze Age contexts in NW Wales.


Opportunities for primary research certainly exists, but in the context of mining sites proof of Roman working is most likely to emerge as a result of ‘tidying-up’ activities revealing datable artefacts on sites of post-Roman date.


 ‘Normal agricultural activity’ is damaging the site at Holt, a site that is currently the subject of on-going programme of collaborative research under the auspices of the Grosvenor Museum, Chester.  A programme of geophysical survey should lead to a greater understanding of this complex.


Geophysical survey in the vicinity of auxiliary forts may lead to the detection of pottery and tile/brick kilns.





The general character of military, military-related, and for want of a better term  ‘ classical Roman’ funerary practices, associated with auxiliary forts and industrial complexes such as Pentre, is reasonably clear.


Several sites have produced examples of earlier Roman cremation burials (military and civilian) together with later inhumations.  A RB landscape and a small cemetery were partly explored under controlled conditions at Ruthin, whilst inhumations dug through vicus deposits at Caersws testify to late Roman changes in funerary practice.


Whilst the presence of some cemeteries can be broadly established, there is a probability that the location of others may be influenced by the road system in the vicinity of forts or industrial complexes.


Several suspected  ‘holy places’ are identifiable in the archaeological record.  Of these two  -Abercynafon and Tre Owen, Newtown- show apparent continuity of watery places as cult foci into the RB period.



The evidence for funerary practices is heavily biased towards military establishments.  Virtually nothing is known of funerary custom among the native community. Is the remarkable Smithfield (Welshpool) deposit in any way typical of rites practised by the elite?


We do not know to what extent pre-Roman funerary practices continue into the RB period, and those of the Roman age into the Early Christian period.    It is tempting to see the Tandderwen-type necropolis as originating within the RB period.  In this respect we may note the grave surrounded by a square enclosure/? roofed building at Ruthin in an apparently early Roman context.


We have scant information on those buried in ‘military’ cemeteries, and on the differences in burial rite and grave type, from ‘bustum’ –type graves to built tombs.  Only three tombstones are known, all from one site, Brecon Gaer.


No certain temples/shrines are known in a military context, let alone outside the military sphere. No altars/dedications to illuminate cult activity and examples of syncretism are known.


There has been no attempt to certainly establish the existence of a built shrine in this period.  The testing of the square enclosure  -possibly a temenos - close to Forden Gaer, with inconclusive results, is about as far as research has gone.  It is conceivable that geophysical survey may throw light on these enigmatic viereckschanzen.


We know nothing about the early Christian community  -if such existed- in the region.



The possibility that work on multi-period burial sites may throw light on one aspect of native funerary practice in the Roman period.


Geophysical survey to determine the extent of cemetery areas in the vicinity of auxiliary forts.  Similarly, examination of viereckschanze – type enclosures in order to determine whether some, or all, may be cult foci.



A lack of interest in the subject.  The relatively invisible nature of much of the subject matter, and the absence of built shrines inhibits a scholarly input.




Where do we want to go in the future?  How might we prioritise future research?

Clearly, we cannot address socio-economic issues differently and challenge accepted concepts, as urged by the contributors to James & Millett’s volume, without substantial enhancement of our database embracing many fields.  Indeed, we are still at a basic data-gathering stage in our region.


On the military front the search for and investigation of pre-Flavian installations   -marching-camps, campaign bases and auxiliary forts should continue.  Much also remains unclear as to the chronology of the Flavian and later military occupation, particularly the strength of the auxiliary garrison in the second century and beyond.  The investigation of forts with an apparent third-fourth century occupation is also urgently required.


The Roman Forts Environs Project undertaken by the Gwynedd Archaeological Trust has produced excellent results.  It should be extended to every site in Wales, thereby enabling us to target resources to face any future threat to vici and other elements of these complexes, as well as allowing more effective management of these sites as a whole.


It is imperative that we begin to address the problem of the invisibility of rural settlements of the Roman period in certain parts of our region.  A programme of targeted research excavations seems to be the only way forward here.   The launching of Wroxeter Hinterland-type projects focussing upon the relationship between military and civilian in areas such as those surrounding Caersws or Forden, coupled with a chronologically wider programme of landscape analysis and environmental sampling, is desirable.


The lack of evidence for settlement continuity between the late Roman and Early Christian period needs to be addressed.  The existence of such is hinted at by evidence from sites such as New Pieces.  In this respect certain critical zones, such as the area around Forden, have some potential for establishing long-term occupation sequences.


An acceleration of adequately funded aerial reconnaissance to target key areas where crop-mark formation is likely to produce evidence of both early military sites and settlements of the pre-Roman Iron Age and RB period.  An additional bonus here would be the detection of high-status, stone-built farms, as has occurred in Shropshire.  Cross-border co-operation in respect of such a project is highly desirable.


The interest which the discovery of maritime remains have aroused in SE Wales could be tapped through the investigation of the late prehistoric/RB foreshore, if, as seems likely, the coastline extended further inland than it does to-day.  Since the transport of lead pigs to the Wirral is archaeologically demonstrable, we could well find Roman wharfing facilities on the Flintshire coast.


It is essential that research programmes embracing later prehistory take note of the high probability of a Roman dimension, and that collaborative schemes be undertaken given the fact that our region straddles the border between England and Wales.



Paper prepared by Jeffrey L Davies. University of Wales, Aberystwyth



This document’s copyright is held by contributors and sponsors of the Research Framework for the Archaeology of Wales.